the fiscal impact of tax-credit scholarships in georgia

8/14/2019 The Fiscal Impact of Tax-Credit Scholarships in Georgia

1/36

arent Choice or Georgia:

Many agree with the concept.

ome disagree. And some

mply want more inormation.

s the public debate continues

o grow louder about how best

o provide a quality education

o all Georgia children, it is

ritical to know the acts about

arent choice, and to have an

nderstanding o how parent

hoice programs have had

n impact on communities,

arents and students around

he country. All o this analysis

done with one goal in mind:

he best possible education or

l o Georgias children.

The Fiscal Impact o

Tax-Credit Scholarships

in Georgia

Prepared By:Brian Gottlob

Senior Fellow

Friedman Foundation or Educational Choice

February 2008

Study released jointly by the Friedman Foundation or Educational Choice, Allian

or School Choice, Georgia Public Policy Foundation, Americans or Prosperity

Black Alliance or Educational Options, and Georgia Family Council


2/36

A MESSAGE FROM THE FRIEDMAN FOUNDATION:

OUR CHALLENGE TO YOU

Our research adheres to the highest standards o scientic rigor. We

know that one reason the school choice movement has achieved such

great success is because the empirical evidence really does show that

school choice works. More and more people are dropping their oppo-

sition to school choice as they become amiliar with the large body

o high-quality scientic studies that supports it. Having racked up a

steady record o success through good science, why would we sabotage

our credibility with junk science?

This is our answer to those who say we cant produce credible research

because we arent neutral about school choice. Some people think that

good science can only be produced by researchers who have no opin-

ions about the things they study. Like robots, these neutral researchers

are supposed to carry out their analyses without actually thinking or

caring about the subjects they study.

But whats the point o doing science in the rst place i were never al-

lowed to come to any conclusions? Why would we want to stay neutral

when some policies are solidly proven to work, and others are proven

to ail?

Thats why its oolish to dismiss all the studies showing that school

choice works on grounds that they were conducted by researchers who

think that school choice works. I we take that approach, we would

have to dismiss all the studies showing that smoking causes cancer,

because all o them were conducted by researchers who think that

smoking causes cancer. We would end up rejecting all science across

the board.

The sensible approach is to accept studies that ollow sound scientic

methods, and reject those that dont. Science produces reliable empiri-

cal inormation, not because scientists are devoid o opinions and mo-

tives, but because the rigorous procedural rules o science prevent the

researchers opinions and motives rom determining their results. I

research adheres to scientic standards, its results can be relied upon

no matter who conducted it. I not, then the biases o the researcher

do become relevant, because lack o scientic rigor opens the door or

those biases to aect the results.

So i youre skeptical about our research on school choice, this is our

challenge to you: prove us wrong. Judge our work by scientic stan-

dards and see how it measures up. I you can nd anything in our work

that doesnt ollow sound empirical methods, by all means say so. We

welcome any and all scientic critique o our work. But i you cant nd

anything scientically wrong with it, dont complain that our ndings

cant be true just because were not neutral. That may make a good

sound bite, but what lurks behind it is a fat rejection o science.


3/36

The Fiscal Impact o

Tax-Credit Scholarships

in Georgia

Prepared By:

Brian GottlobSenior Fellow

Friedman Foundation or Educational Choice

February 2008


4/36

State SurveysNevada

January 2008

Illinois

December 2007Georgia

April 2007

Florida

January 2006

Issues in DepthGrading School Choice: Evaluating School Choice Programs by the Friedman Gold Standard

February 2008

Monopoly vs. Markets: The Empirical Evidence on Private Schools and School Choice

October 2007

Disruptive Behavior: An Empirical Evaluation o School Misconduct and Market Accountability

June 2007

Education by the Numbers: The Fiscal Eect o School Choice Programs, 1990-2006May 2007

Freedom rom Racial Barriers: The Empirical Evidence on Vouchers and Segregation

September 2006

Funding School Choice: A Road Map to Tax-Credit Scholarship Programs and Scholarship Granting Organizations

June 2006

Using School Choice: Analyzing How Parents Access Educational Freedom

October 2005

Grading Vouchers: Ranking Americas School Choice Programs

January 2003

School Choice Works! The Case o Sweden

January 2003

Learning rom Success: What Americans Can Learn rom School Choice in Canada

January 2002

Issues in the StateThe High Cost o High School Failure in New Jersey

February 2008

The Fiscal Impact o a Tuition Assistance Grant or Virginias Special Education Students

April 2007

Utah Public Education Funding: The Fiscal Impact o School Choice

January 2007

The High Cost o Failing to Reorm Public Education in Indiana

October 2006

Segregation Levels in Milwaukee Public Schools and the Milwaukee Voucher Program

August 2006

Floridas Public Education Spending

January 2006

Spreading Freedom and Saving Money: The Fiscal Impact o the D.C. Voucher Program

January 2006

The Constitutionality o School Choice in New Hampshire

May 2005

An Analysis o South Carolina per Pupil State Funding

February 2004

A Guide to Understanding State Funding o Arizona Public School Students

January 2004

The Eects o Town Tuitioning in Maine and Vermont

January 2002

For a complete listing o the oundations research please visit our website at www.riedmanoundation.org.


5/36

Executive Summary

This study seeks to inorm the debate over a proposal in Georgia to give tax credits or contributions to organizations

that provide scholarships to K-12 private schools. Such a program would extend to K-12 education the philosophy o

choice that is inherent in the states existing Hope Scholarships program or college students. This study constructs a

model to determine the scal impact o tax-credit scholarships on the state and on local school districts.

In addition to allowing Georgia to expand educational opportunity and improve the equity o its education system

a tax-credit scholarship program would generate large scal benets or local school districts, increasing the available

resources or students who remain in public schools. Because much o their revenue does not vary with enrollment,

school districts would retain much o the unding associated with students who use scholarships to migrate rom public

to private schools. The overall impact on public schools would be to increase the nancial resources available per

student. Depending on how the program is designed, it could also result in a scal savings to the state budget. Even i

it does not produce a savings at the state level, costs to the state would be signicantly reduced because public schoo

spending would go down as students leave with scholarships.

Key ndings include:

When students leave Georgia public schools, their local school districts experience reductions in expenses that

are greater on average than the reduction in their state aid. In addition, school district revenues rom local

sources do not decline when enrollments decline. Because expenses decline more than revenues when students

leave public schools, private school scholarships produce a net gain o resources available to students who

remain in public schools. The gain is equal to about $6,600 per public school student using scholarships.

The total scal impact o a tax-credit scholarship program depends on the number o public school students

who are induced to participate. This in turn depends on a number o program design actors, such as income

eligibility levels, the size o the scholarships and the total amount o available scholarship unding. The study

uses data rom the Census and other sources to estimate how public school amilies might respond to a tax

credit scholarship program with various design eatures.

A scholarship program or current public school students that provided $50 million in scholarship unding;

established income eligibility at 200 percent o the ree and reduced-price lunch level; and awarded scholarships

o $3,500 each would produce a total scal benet o $94 million or local school districts.

The impact o the program on the state budget depends on program design eatures as well. For example,

a program with the design eatures in the example above would produce a savings o almost $6 million or

the state.

Raising the income eligibility or scholarships always increases the scal benet o the program, because

more public school students would be eligible or scholarships and eligibility is increased most among income

groups that have the highest propensity to migrate rom public to private schools.

A tax-credit scholarship program is a more ecient way to direct dollars to education than increasing state

aid. Georgia data show that every dollar o increased state aid to schools only produces an additional 53

cents o additional school spending, because local governments respond to the state spending increase by

reducing local spending on education. By contrast, every dollar spent on a tax-credit scholarship program is

a ull dollar that goes to education.


6/36

About the Georgia Public Policy FoundationSince 1991, the Georgia Public Policy Foundation has conducted scholarly research and analysis o state

public policy issues and worked to educate citizens, policy-makers and the media. The 501(c)(3) is state-

ocused, independent, non-partisan and market-oriented in its approach. Its philosophy is that good public

policy is based upon act, an understanding o sound economic principles and the core principles o our reeenterprise systemeconomic reedom, limited government, personal responsibility, individual initiative,

respect or private property and the rule o law.

About the Friedman Foundation or Educational ChoiceThe Friedman Foundation or Educational, dubbed the nations leading voucher advocates by the Wall Street

Journal, is a nonprot organization established in 1996. The origins o the oundation lie in the Friedmans

long-standing concern about the serious deciencies in Americas elementary and secondary public schools.

The best way to improve the quality o education, they believe, is to enable all parents with the reedom to

choose the schools that their children attend. The Friedman Foundation builds upon this vision, claries its

meaning to the public and amplies the national call or true education reorm through school choice.

About the AuthorBrian J. Gottlob ([email protected]) is the principal o PolEcon Research. For 17 years Gottlob has analyzed eco-

nomic, demographic, labor market industry and public policy trends or private sector, government and nonprot organizations.

He has extensive experience in developing econometric models and has completed studies on a range o economic, tax policy,

energy, education, and health care issues in New Hampshire, Virginia, Ohio, New Mexico, New York, Texas, Oregon, Michigan,

Georgia, Mississippi, West Virginia and Illinois. Gottlob is a senior ellow at the Friedman Foundation or Educational Choice.

He has been an instructor at the Whittemore School o Business and Economics at the University New Hampshire, a member

o the Advisory Board o the New England Economic Partnership and a member o the National Association o Business Eco-

nomics. Prior to ounding PolEcon, Gottlob was a Vice President or Fiscal and Economic Policy at the Business and Industry

Association o New Hampshire. He has an undergraduate degree in economics rom the State University o New York and a

graduate degree in public policy analysis rom the University o New Hampshire.

About Black Alliance or Educational OptionsThe Black Alliance or Educational Options (BAEO) is a national, nonprot, nonpartisan membership organiza

tion whose mission is to actively support parental choice to empower amilies and increase quality educationa

options or Black children. Staunch in its belie that parental choice must be an integral part o any serious

eort to reorm education in America, BAEO believes parental choice programs, which lead to the creation

o quality educational options, not only rescue the children who can take advantage o such opportunities bu

also create powerul incentives or all schools, public and private, to improve.

About the Georgia Family CouncilGeorgia Family Council (GFC) is a non-prot organization that exists to strengthen and deend the amily

in Georgia by equipping marriage advocates, shaping laws, preparing the next generation and infuencing

culture. GFC is working with leaders throughout the state to prevent amily breakdown and its costly and

devastating eects on the lives o children, adults, businesses and communities.

About the Alliance or School ChoiceThe Alliance or School Choice works to build support or and implement publicly unded school choice

programs that provide low-income amilies with educational opportunity. In doing so, the Alliance not only

protects those programs that are already serving amilies in need, but also expands and enhances them

andmost importantlyinitiates new, larger and even more eective models.

About the Americans or ProsperityAmericans or Prosperity (AFP) and Americans or Prosperity Foundation (AFP Foundation) are committed

to educating citizens about economic policy and mobilizing those citizens as advocates in the public policy

process. AFP is an organization o grassroots leaders who engage citizens in the name o limited government

and ree markets on the local, state and ederal levels. The grassroots members o AFP advocate or public

policies that champion the principles o entrepreneurship and scal and regulatory restraint


7/36

Table o Contents

9

10

10

12

13

15

15

16

17

19

21

21

22

23

27

28

30

33

33

34

Introduction

How Georgia Funds Public Schools

How QBE Works

How School Funding Varies with Enrollment

How School Expenditures Vary with Enrollment

How State Education Aid Aects Local Education Expenditures

Each Dollar o Additional State Aid Increases School Expenditures by Only 53 Cents

Demographics o Georgias School-Age Children

Characteristics o Public and Private School Children in Georgia

Proposals to Increase Educational Options and Opportunities

Estimating Program Participation Levels

Business Contributions to Scholarship Organizations

Scholarship Contributions rom Individuals

Tuition Prices Strongly Infuence Demand or Private Schools

Combining Supply and Demand Models to Estimate the Number o Scholarships

Program Tradeos

Fiscal Impact on the State

Fiscal Impact on Local School Districts

Conclusion

Endnotes


8/36

The Fiscal Impact o Tax-Credit Scholarships in Georgia

February


9/36


February 2008

Introduction

Proposals to increase educational opportunities and choices or students o dierent backgrounds, abilities, needs and

economic circumstances are increasing throughout the country. In part, this refects increasing support among the public

or the concept o school choice; a majority o U.S. citizens now support it.1

In Georgia, substantial majorities avor school choice. A 2007 opinion poll ound that 58 percent o likely voters in Georgia

supported school vouchers, while only 22 percent opposed them. Four times as many said they would be more likely to vote or

a state representative or senator who supported vouchers as said they would be less likely (54 percent versus 13 percent).2

The philosophy o school choice is already inherent in Georgias Hope Scholarship program or college students. Rather

than reserving state support or students who attend colleges assigned to them by the state, Georgia provides support or

students to attend the public or private colleges o their choice. The Hope Scholarship program is not limited by income,

refecting a decision on the part o the state that school choice is not just good or poor students, but or students o all

demographic backgrounds.

In K-12 education, amilies in Georgia already exercise school choice in several ways:

According to U.S. Census Bureau data, about 128,000 school-age children in Georgia attend private schools.3

About 36,000 Georgia students are home-schooled.

About 26,000 Georgia students attend charter schools.

From 1997 to 2006, home schooling nearly doubled (a 93 percent increase) in Georgia, while private school enrollment

increased 25 percent and public school enrollment increased 16 percent. Since 2004, charter school enrollment in Georgia has

increased rom 14,828 to 26,299 (a 77 percent increase).

But by ar the major orm o choice in Georgia and in the United States occurs when a amily chooses its place oresidence. Because higher-quality schools oten are ound in communities with higher housing prices, this type o school choice

is available at a high cost, and is unaordable to many amilies. Throughout the United States, afuent amilies exercise a lot

o choice. Georgia amilies sort themselves among schools and school districts on the basis o parents income and education.

But i education unding were made available to parents, in order to help amilies have more choices among schools, we

would see less segregation on the basis o parental income and education. The positive response by amilies in Georgia over

the past several years to a wider variety o educational options and choices suggests that Georgia should consider the merits

o providing even more educational options or school children.

This study uses empirical methods and the tools o economic analysis to examine school choice in Georgia. Empirical

analyses allow us to nd analytical answers to important policy questions outside the hardened positions o support or opposition

that come rom viewing school choice simply as a matter o political leanings or ideological principle. Perhaps believingit inappropriate to discuss education in terms o market incentives and pressures, many well-meaning individuals who are

deeply concerned about K-12 education ignore, on principle, the educational impacts o market orces and how they infuence

the behaviors o amilies and schools. But school choice occurs even in the absence o ocial or legislatively enacted school

choice policies. Unortunately, the market or K-12 education without universal school choice contains signicant imperections

that prevent many amilies rom being able to send their children to the schools that meet their needs.

Our analysis begins with a brie discussion o how Georgia unds elementary and secondary education. We examine

the demographics o public and private schools in Georgia and estimate the impact on public and private school enrollments

o a program that provides tax credits or donations to support private school scholarships. We develop a model that shows


10/36


0 February

how the expenditures o Georgia school districts vary with changes in student enrollment, and show the scal impacts o a

school choice program on Georgias state budgets and those o local school districts.

How Georgia Funds Public Schools

The expense o educating children in Georgia, as in most states, is a responsibility shared between the state and local

governments, with the state providing about hal the unds used to educate children in local public schools. In 1985, Georgia

enacted the Quality Basic Education Act (QBE). Prior to the QBE, the state provided education unds to local districts simply

on the basis o the number o students enrolled, without any adjustment or the scal resources o each district. Perceived

inequalities o education unding among school districts were the driving orce behind adoption o the QBE. The QBE nance

program is a oundation aid ormula designed to equalize the nancial resources that each school district has, at a specic

tax rate, to meet the cost o providing a basic level o educational services to all public school students in a district.

Funds or capital outlays (school construction) and debt service also are provided by the state, though not through

the QBE. Including these unds would require a more complex analysis than can be provided in this study, so they have

been excluded.

There are three components o the QBE that distribute state unds to school districts, they are:

QBE earnings

Categorical grants

Equalization

How QBE Works

1) QBE Earnings

The amount o unding each district receives or QBE earnings is determined by the ollowing ormula:

QBE earnings = (Direct Instructional Costs + Indirect Instructional Costs) Local Share

Direct instructional costs are the costs o putting a teacher in every classroom. The cost or each student is weighted based

upon the grade level o the student (K, 1-3, 4-5, 6-8, or 9-12), any special program in which the student is enrolled (Early

Intervention Program, special education, gited, remedial, alternative, middle school, English as a Second Language, or

vocational lab), and the established teacher-student unding ratio. These weights refect the expected costs associated with

educating students at dierent grade levels and with dierent needs and abilities. The benchmark weight o 1.00 is assigned to

a student in grades 9-12 who attends regular classes and receives no special servicesthe least expensive student to educate.

All other students receive a higher weight. The resulting weight or each student is known as the Full Time Equivalent

(FTE) count. The FTE count used in the QBE ormula is always higher than the actual number o students.4

Indirect instructional costs include unding or district administration, school administration, sta and proessional

development, acilities maintenance and operations, media centers and 20 additional days o instruction. These costs are

calculated based on system size, school size or student population.

Local share reers to the portion o direct and indirect instructional costs that Georgia expects school districts to pay

with locally raised unds. Each district is required to raise school unding through local taxes, with the minimum amount


11/36


February 2008

equal to what would be raised by a property tax o at least 5 mils, where one mil represents a $1 tax on each $1,000 o real

property valuation. By law, local unds should provide no more than 20 percent o the total amount required to provide a

basic education to all Georgia public school students, but or a number o reasons the percentage has in act exceeded

that amount.5 In addition to the required local share o 5 mils, school districts may levy up to another 15 mils without asking

or voter approval.

2) Categorical GrantsLocal districts receive state education unds or certain non-instructional purposes such as student transportation,

sparsity, principal support and certain low-incidence special education needs. The amount o categorical grants is small:

about $171 million in 2007-08.

3) Equalization

QBE earnings provide unds or a basic, adequate education. Equalization is an attempt to address the issue o equity.

Because o dierences in assessed property wealth, local school districts dier in their ability to raise revenue at similar

tax rates. Georgia attempts to overcome some o this inequality by making it easier or some local school districts to und

schools at lower property tax rates. Equalization unds are allocated to districts with per-student taxable property wealth

that ranks below the 75th percentile o all districts. Equalization provides an incentive or districts to raise local property

taxes by tying the amount o equalization unding to the property tax level. Districts at or below the 75th percentile levelcan receive equalization unding in proportion to the amount o mils they levy beyond 5 mils. In 2007-08, the total amount o

equalization unding was about $458 million.

Each school districts basic program entitlement under QBE is determined by multiplying the weighted FTE pupil

count by the guaranteed nancial support base, which is set by the state legislature. In 2007-08 the guaranteed nancial

base was $2,642 per weighted FTE. The base amount represents the costs associated with educating a grade 9-12 student who

is in no special programs. The base amount is apportioned as ollows: 79.5 percent or direct instruction (salary, benets,

textbooks, supplies, personnel travel, replacement equipment, etc.); 17 percent or indirect costs (maintenance and operations,

support personnel, school and central oce personnel, etc.); 0.8 percent or sta development; and 2.8 percent or media

costs (personnel, materials and equipment). Ater subtracting the amount raised by the required local share, the remaining

cost o providing a basic education is provided to districts as QBE earnings.

STATE SOURCES

Funding Based on Enrollment

Categorical Aid and Equalization

TOTAL STATE SOURCES

LOCAL SOURCES

FEDERAL SOURCES

TOTAL*

PercentPer Student (FTE)

$6,468,897,783

$659,902,299

$7,128,800,082

$5,767,032,930

$1,003,997,700

$13,899,830,712

Amount

Georgia Public School Revenue 2006-07(Excludes capital projects, debt service and some other categories)

Table 1

$3,931

$421

$4,352

$3,603

$627

$8,582

90.3%

9.7%

50.7%

42.0%

7.3%

100%

Source: Earnings Sheet for 2007, Georgia Department of Education District Revenue and Expenditure Reports, and author calculatio

* The Earnings Sheet for 2007 does not provide data on all school spending categories. The Capital Projects Fund, the Debt Service Fund and a number of other categories are exclud


12/36


2 February

How School Funding Varies with Enrollment

The relationship between enrollment levels and school unding is a question o particular importance or determining

the scal impact o school choice programs. Funding rom dierent sources responds to changes in enrollment in dierent

ways. While most school unding that comes rom the state varies with enrollment, local school unding does not.

In 2007-08, 90 percent o state operating support or public schools (which excludes such categories as construction

unds and debt service unds) is calculated on the basis o enrollment. It thus varies as a district adds or loses students, and

also according to the weights assigned to each student. Categorical aid and equalization aid (a combined $630 million) and

unds or school nurses ($30 million) do not vary directly with enrollment.7 Because a high percentage o state education

unding in Georgia is distributed without the restrictions that accompany categorical or narrow-purpose grants, Georgias

unding ormula provides a good deal o spending fexibility or local districts. Although more than 72 percent o QBE earnings

unds allocated to each district are allocated or direct instructional purposes (the costs o putting teachers in classrooms)

and are required to be used or that purpose, this requirement is not very restrictive because classroom instruction is the

primary unction o the school system. The remainder o state QBE earnings unding is allocated or indirect costs, including

administrative, support and maintenance expenses. The ungibility o money makes it easy or school districts to satisy

these requirements.8

Taxes on real property are the primary source o local revenue or school districts. These revenues are determined

by local property values and tax rates. In the short term, enrollment does not aect local revenue. Over time, it is possible

that enrollment changes may prompt adjustments to local revenues, but more oten local revenues do not decline even

as enrollments do.

Some revenues rom ederal sources vary with enrollment, such as unds or the reimbursement o ree and reduced-

price lunch, breakast and milk programs. Since ederal unding or schools is dispersed across a large number o unding

streams, it is dicult to determine the exact percentage o ederal unding that varies with enrollment.

Table 1 uses 2006-07 unding levels as reported by the Georgia Department o Education. The table shows that, or

each student, the state provides an average o about $4,352 in education aid (excluding capital unds, debt service unds,

etc.). O the $4,352, about 90 percent, or $3,931, is directly responsive to changes in enrollment levels. Thus when a new

student enters a school district, the district receives $3,931 in additional state unding on average. Conversely, when a

student leaves a district, state per-student unding is reduced by $3,931 on average, with the district retaining $421 in state

STATE SOURCES

Funding Based on Enrollment

Categorical Aid and Equalization

TOTAL STATE SOURCES

LOCAL SOURCES

FEDERAL SOURCES

TOTAL

$5,896,500,000

$659,902,299

$6,556,402,299

$5,767,032,930$1,003,997,700

$13,327,432,929

Amount

Change in Georgia School District Revenues Resulting rom a Decline in Enrollment(Enrollment Decline From 1,567,486 to 1,500,000)

Table 2

Per Student

(FTE)

$3,931

$440

$4,371

$3,845

$669

$8,885

Change Per

Student

$0

+$19

+$19

+$242

+$42

+$303

Revenue Source

-$572,397,783

$0

-$572,397,783

$0

$0

-$572,397,783

Change rom Actual

2006-07 Revenue


13/36


February 2008

unds, the entire $3,603 rom local sources and much o the $627 rom ederal sources. In the long run, all revenue is at

least potentially variable with enrollment, with the exact extent dependent upon the decisions o local school boards and

these who approve their budgets. However, these gures give an accurate picture o how revenue changes with enrollment

in the shorter term.

Table 2 urther illustrates how school revenues are aected by enrollment declines. The table shows how Georgias

aggregate and per-student school district revenues would have been aected i enrollment in the 2006-07 school year hadbeen exactly 1.5 million rather than the actual gure o 1,567,486 (a decline o 67,486).

The table shows that, compared to actual revenues or the 2006-07 school year, the decline o 67,486 students would

lower total district revenues by $572.4 million, but per-student revenues would actually increase by $303 per student. Even

in an unrealistic scenario where all ederal and state categorical revenues varied directly with enrollment, per-student

revenues would still increase by $242 because local unding does not vary with enrollment.

The implications o this analysis are:

Under the current system o Georgia public school unding, on average a decline in local district student

enrollments, while resulting in a decline in total revenues, actually produces an increase in the resources availableor educating each student who remains in the district.

Because local school unding does not vary with enrollment, a loss o students cannot result in lower per-student

revenues being available to school districts.

Smaller public school enrollments can result in large savings or the state without reducing the per-student

revenues available to local school districts.

Georgia could, as an alternative to realizing savings rom enrollment declines, choose to use some or all o the

savings in education aid to und other educational programs or education reorm initiatives.

As long as the revenue loss associated with each student who leaves a school district is lower than the amount bywhich total school district expenditures are reduced when a student leaves, a local school district cannot be made worse

o nancially by the loss o a student. In Georgia, the loss o $3,931 on average in state unds, along with perhaps a small

amount o ederal unds, is lower than the expenditures attributable to each child. In the next section we will consider the

extent to which these expenditures vary with enrollment.

How School Expenditures Vary with Enrollment

Evaluating the scal impact o enrollment changes on Georgia school districts requires not only an understanding o

how state education aid to communities is aected, but also some estimate o how expenditures o school districts change

in response to enrollment changes.

When students leave a school district, the district loses state aid associated with those children, but expenses associated

with educating those children also decline. One criticism o school choice is that the loss o students is not accompanied by

a concomitant decrease in expenses. While that may be true in the very short term (less than one school year) or with very

small enrollment changes, the conclusion that expenditures can never decline when enrollments drop produces logically

and empirically implausible conclusions. Increasingly, studies have demonstrated that local school district expenditures

are sensitive to declines in enrollment.9


14/36


4 February

Using detailed school district data rom the Georgia Department o Education, we employed an econometric approach to

estimate the variable expenditures o educating each student in Georgia. We used detailed school district nancial data rom

all districts or the 1999-2000 and 2006-07 school years to determine to what extent operating expenditures (all expenditures

less capital expenditures and debt service) are variable (that is, responsive to the addition or loss o students in a district)

and to what extent they are xed. For this study we considered variable expenditures to be expenses that are variable over

a period o at least a year. This analysis will test the expectation that school districts can and do adjust their expenditures

to refect enrollment levels rom one year to the next.

We developed simple linear regression models to estimate the expenditure structure o public schools in Georgia. Ater

testing many models, we identied the model with the strongest ability to describe and predict the expenditures o public

schools according to changes in enrollment. It is expressed by the ollowing equation:

Where:

This model estimates that the variable expenditures associated with educating each additional public school student

in Georgia in 2006-07 school year were $6,299 ($5,039 in real, infation-adjusted 1999 dollars) or 73 percent o the $8,642

mean expenditure per student across all districts.10 The model explains 98 percent o the nominal change in school district

expenditures between the 1999-00 and 2006-07 school years. The data or this analysis are at the school district level, and, as

a result, enrollment changes over the years examined are much larger than those that occurred in individual schools within

districts. These results are not meant to imply that an increase or decline in a small number o students in a school would

necessarily lead to increases or decreases in school expenditures o $6,299 per student. But in larger numbers and across

districts, over time expenditures are highly responsive to enrollment changes.

a Dependent Variable: ChngExpenditu

0.988

RStandard Error

o the Estimate

8347825

Variable Expenditure Model Summary

Table 3

Adjusted R Square

0.9750.975

R Square

1

Model

Variable Expenditure Model Summary

0.988

RStandard Error

o the Estimate

8347825

Adjusted R Square

0.9750.975

R Square

1

Model

(Constant)

ChgEnrollment

PPWealth

1999Exp

Beta

0.392

0.007

0.721

Std. Error

1486747.200

225.948

12.314

0.010

631413.8

6299.1

6.951

0.473

B

1

Model

UnstandardizedCoecients

StandardizedCoecients

0.425

27.878

0.564

49.796

t Sig.

0.672

0.000

0.573

0.000

Collinearity Statistics

Tolerance

0.722

0.926

0.683

VIF

1.384

1.080

1.464


15/36


February 2008

Because the change in expenditures associated with each student who enters or leaves Georgias public schools is

greater than the state education aid per student, the loss o students rom a school district would have a net positive impact

on local school district nances. In the 2006-07 school year, the loss o a student rom a district would mean the loss o about

38 percent o revenues associated with that student (about $3,931 in state education aid), leaving another $4,651 in per-

student revenues in the district. At the same time, the district would (on average) see a decrease in expenditures o $6,299.

Thus, at least in the short run, school districts are nancially better o. Even i enrollment declines resulted in the loss o

all categorical state aid (an unrealistic scenario) and all ederal aid (similarly unrealistic), each district still would retain$3,603 per student in local revenues.

Each year thousands o school children migrate between school districts; individual districts lose and gain students;

and local districts regularly adjust their expenditures to accommodate these changes. Our analysis o U.S. Census Bureau

data suggests that about 12 percent o Georgia school age children in 2007 live in a dierent home or apartment than they did

in 2006.11 This implies that about 170,000 public school students change residences each year, and it is likely that a signicant

percentage o those who move do so in a way that requires changing school districts. The number is likely to ar exceed the

number who would participate in a tax-credit scholarship program. Our analysis shows that concerns over the potential

scal impacts o school choice on local school districts not only are overstated, but they ail to understand the undamental

local district scal eect o expanding school choice in Georgia: an increase in the resources available or each student

who remains in the school district.

How State Education Aid Aects Local Education Expenditures

Developing a time-series database o state and local education nance variables in Georgia allows us to examine and

understand the impacts o enrollment changes and other important issues in education nance. There are at least three

important and related education nance questions that may infuence an evaluation o any program to expand the education

options available to Georgia amilies:

How does state education aid aect the aggregate level o local expenditures?

How does state education aid aect dierent expenditure categories?

How does state education aid aect the equitable provision o education services across local districts and among

children across and within districts?

None o these questions is the central ocus o this study, but thoughtul consideration o a tax-credit scholarship

program cannot escape these critical questions. Consideration o equity across local districts and among individual children

is a topic or another study.

Each Dollar o Additional State Aid Increases School Expenditures by Only 53 Cents

Increases in state education aid do not necessarily result in a concomitant increase in educational expenditures by

school districts. Local school districts can respond to an increase in state aid by reducing the local tax burden or education,

or by shiting local expenditures rom education to non-educational categories without either reducing tax collections or

increasing education expenditures.12

Based on our analysis o Georgia spending data, we estimate that between 1999 and 2007 each additional dollar o state

education aid resulted in only 53 cents o additional education spending in public schools. This gure is very close to estimates

nationally and is actually well above the 40 cents ound in a recent study o New Hampshire education nance.13 Nevertheless,

it does oer a cautionary note to those who seek to increase education expenditures by increasing state aid to schools.


16/36


6 February

The legal requirement that state education aid be used or education spending is easy to meet without actually increasing

local education budgets. Local governments can simply move the same number o local tax dollars out o the school budget

as the number o state dollars that come into the school budget. A local government that reduces education expenditures by

$1 million when it accepts an additional $1 million in state education aid has complied with the legal requirement to spend

state education aid on education. The local government could spend the $1 million in new state unding on education while

spending $1 million in local tax revenuewhich it would have raised or education in the absence o the state aid or other

local services, in which case education spending would not increase and taxes would not be lowered, but spending on otherlocal services would increase. Or it could simply reduce local taxes by $1 million, in which case taxes would be lowered

while spending on both education and other local services would not increase. As our results and those o other researchers

indicate, the most likely outcome o additional state education aid is a combination o increases in education spending,

increases in non-education spending increases, and tax relie.

The results o this research also indicate that the amount o additional state education aid used or increased education

expenditures (53 cents) is divided approximately according to state regulations, with about 72 percent (or 38 cents) going

or instructional purposes and about 28 percent (15 cents) going to increase non-instructional expenditures.

These ndings and those in prior sections o this study support at least three important conclusions that are relevant

to an evaluation o a tax-credit scholarship program:

Because each dollar o additional state education aid, on average, translates into about 53 cents o additional

local spending on education services, the current system o education nance is a relatively inecient method o

increasing educational services and educational opportunities or Georgia students.

By contrast, education unding that provides scholarships or students to attend schools would result in $1 o

educational expenditures or each $1 o unding (i administrative expenses or the program are counted as educational

expenditures, as are administrative expenses in the public school system). For those most concerned with creating

equitable educational opportunities across schools and districts, tax-credit scholarships are thus a more ecient

mechanism or directing expenditures to education and or providing increased educational opportunities than is

increased state education aid.

Because the reduction in school district revenue associated with declines in enrollment is less than variable cost o

educating students, school districts cannot be made nancially worse o (over periods o more than one year) by

the loss o students to a scholarship program. This point will be discussed in greater detail later in this study.

Demographics o Georgias School-Age Children

Forty percent o students in Georgias public schools qualiy or the ederal ree and reduced-price lunch program.

That means they reside in households with income levels at or below 185 percent o ederal poverty guidelines, which are

based on amily size. A amily o our can earn up to $39,220 in 2008 and have its children qualiy or ree and reduced-pricelunches (see Table 4).

The percentage o students eligible or ree and reduced-price lunch reported here is lower than that reported by the

Georgia Department o Education. It reports that about 841,000 public school children qualied or ree and reduced-price

lunches in 2007.14 Because our analysis examined both public and private school students and looked at some detailed

demographic actors or which the Georgia Department o Education does not make data available, we used U.S. Census

Bureau 2006 data or our estimates o the characteristics o Georgia school children. In addition, we limited the age range

o students or our analysis to ages 5-17. Our results indicate that 639,115 students ages 5-17 in K-12 public schools qualiy or

ree or reduced price lunches and another 22,000 students qualiy in private schools (see Figure 1).


17/36


February 2008

Characteristics o Public and Private School Children in Georgia

According to the Georgia Department o Education, about 8 percent o Georgia school-age children either attend private

schools or are home schooled. Local districts are required to report the number o private school students in their districts.

However, the Department o Education does not veriy the numbers, and school districts ability to produce accurate counts,

and their incentives or doing so, are uncertain. Our analysis o Census Bureau American Community Survey data, 2000

2008 Poverty and Free/Reduced Lunch Income Guidelines

Table 4

$10,400

$14,000

$17,600

$21,200

$24,800

$28,400

$32,000

$57,720

$77,700

$97,680

$117,660

$137,640

$157,620

$177,600

$48,100

$64,750

$81,400

$98,050

$114,700

$131,350

$148,000

$38,480

$51,800

$65,120

$78,440

$91,760

$105,080

$118,400

$19,240

$25,900

$32,560

$39,220

$45,880

$52,540

$59,200

1

2

3

4

5

6

7

Family Size

Free and

Reduced Lunch

Eligibility

200% 250% 300%Poverty

% o Free and Reduced Lunch

0 500,000 900,000 1,300,000 1,800,000

556%+

464-555%

371-463%

186-370%

Up to 185%

Author analysis of U.S. Census Bureau American Community Survey 2006 data for Georgia

Forty Percent o Georgias Public School Students Qualiy or the

Federal Free and Reduced Lunch Program

Figure 1

CUMULATIVE NUMBER OF GEORGIA STUDENTS, 2006

FAMILYINCOMEASA%O

FPO

VERTY

[Free/Reduced Lunch Eligible]

639,115

[200% Free/Reduced Lunch Eligible]

1,151,043


1,300,488


1,383,160

[More than 300% Free/Reduced Lunch Eligible]

1,589,839

21,934

66,126

84,901

110,432

164,911

Public Schools

Private Schools


18/36


8 February

Decennial Census data, and Census Bureau Current Population Survey data suggests that about 10 percent o K-12 students,

or nearly 165,000 students, are enrolled in private schools or home schooled.

In the absence o a universal school choice program, parents will largely exercise school choice by choosing to live in

communities that best match their preerences or educational services or by paying to have their children attend private

schools. One result o the absence o a universal choice program is that amilies and school children segregate themselves

along lines o income, parental educational attainment and race and ethnicity.

This segregation is apparent in Georgia. An examination o the characteristics o Georgia school children provides

some indications o the tendency to segregate in the absence o school choice.

The percentage o children in private schools in Georgia is relatively low, but among white students the rate is higher

than the U.S. average. Compared to the U.S., a smaller percentage o Hispanic and Arican-American children in

Georgia attend private schools (see Figure 2).

A quarter o students in Georgia public schools come rom amilies with an annual income below $25,000, comparedto just 8 percent o students in private schools. Figure 3 shows the income distribution o both public and private

school students in Georgia.

Children in Georgia are more segregated by income than children in most states. Compared to the U.S. average, lower-

and middle-income amilies in Georgia have lower private school enrollments while, among amilies with higher incomes,

private school enrollments are as high or higher in Georgia than in the U.S. Nevertheless, the demand or private schooling in

Georgia increases signicantly as amily income increases, suggesting an income elasticity o demand or private schooling

o approximately 0.5 at lower income levels and 1.6 at the highest income levels. Income elasticity reers to the change in

demand or private schooling that occurs with each percentage-point change in amily income. An elasticity o 0.5 indicates

25%

20%

15%

10%

5%

0%

Georgias Minority Children Have Fewer Educational Choices

Figure 2

13.4%

14.8%

6.7%

3.9%

5.6%

White Arican-American Hispanic

Georgia

U.S.

Author analysis of U.S. Census American Community Survey 2006 data for Georgia

4.1%


19/36


February 2008

that, i amily income doubled (an increase o 100 percent), there would be a corresponding increase in private school

attendance o 50 percent. Figure 4 shows that lower and lower-middle income amilies in Georgia are underrepresented in

private schools compared to the U.S. average.

Together, these data suggest that:

There are substantial economic and racial dierences in the composition o public versus private schools in Georgia,

indicating a dierence in the ability o parents to choose private schools or their children.

The rates o private school enrollment among Georgia amilies o middle through higher incomes compared to

enrollment among lower-income amilies suggest that a large percentage o Georgians view the public schools as

a less attractive option or educating their children and amilies and that amily income strongly infuences the

ability o amilies to exercise their preerence or educational services.

Without increased eorts to introduce more school choice programs, the signicant segregation along income and

other lines that is apparent in Georgia schools will likely continue.

Proposals to Increase Educational Options and Opportunities

Along with economic, demographic and other actors, the perceived quality o public schools infuences the demand

or private schooling in a state and a community. Our review o the demographics o Georgias public and private schools

suggests that there is more separation o students along income and racial lines in Georgia than in most states. At the

same time the demand or private schooling by lower-income and minority students likely is not satised, largely because

o the income constraints these amilies are more likely to ace.

30%

25%

20%

15%

10%

0%

Georgia Students Are Segregated by Income(Through Parental Housing Choice or Ability to Choose Schools)

Figure 3

Public Schools

Private Schools

Author analysis of U.S. Census Bureau American Community Survey 2006 data for Georgia


20/36


0 February

Georgia could achieve a number o important scal and educational objectives by increasing the options parents have

or educating their children. Tax-credit scholarships are one method o increasing the options parents have or educating

their children. Proposals or such scholarships have arisen, in part, in response to concerns about the quality o public

schooling and the rising demand or private schooling created by those concerns.

Georgia is considering proposals to allow a tax credit to individuals and business or contributions made to organizations

that provide tuition scholarships to amilies who want to attend private school. Under one proposal, tax credits are cappedat $500 or a single taxpayer or $1,000 or a married couple ling a joint tax return. Businesses could claim a credit against

their corporate taxes up to 75 percent o their corporate tax liability. Total tax credits available to corporate and individual

contributors to student scholarship granting organizations and educational improvement organizations would be capped at

$50 million in the rst year.

30%

25%

20%

15%

10%

0%

Income Makes More Dierence to Educational Choices in

Georgia than in the U.S. as a Whole

Figure 4

Georgia

U.S.

Source: Author analysis of U.S. Census Bureau American Community Survey 2006 data


21/36


February 2008

Forecasting the impact o Georgias proposed tax-credit scholarship program requires that we predict how parents would

respond to the availability o scholarships. To estimate the number o students who would receive scholarships and attend private

schools, we examined the size o the school-age population in public and private schools; the characteristics and dierences o

the population o school children in public and private schools; and how those dierences likely would aect the demand or

scholarships. We analyzed the interactive eects between the volume o scholarship unds available; the average dollar value o

individual scholarship awards; the total number o scholarship awards; and the impact the migration o public school students

to private schools would have on public school enrollments and nances in Georgia.

Estimating Program Participation Levels

The scal impact o the Georgia tax-credit scholarship proposal rst would depend on the volume o contributions and

the value o the tax credits claimed by individuals and corporations that und scholarships. The proposed program calls or a

maximum o $50 million in tax credits available or businesses and individual taxpayers. To begin our analysis, we consider the

volume o contributions we can expect to be claimed.

With the proposed tax credit, businesses and individuals can choose to pay taxes to be used or general state services or

they can contribute to a scholarship granting organization to provide scholarships or students enrolling in private schools or out-

o-district public schools. When businesses or individuals make a contribution to the tax credit program they directly target the

use o their tax dollars to support education. Given this choice, many businesses and individuals can be expected to contribute to

the program. With the proposed tax credit, Georgia increases educational expenditures in a way that does not occur when state

education aid is increased. As we noted earlier, each additional dollar o state education aid increases school expenditures by only

53 cents. With a tax-credit scholarship program, unding acilitated by tax credits would result in $1 o educational expenditures

or each $1 spent. For many business and individuals, the ability to target their unding to educational expenditures would be an

attractive option.

Several states oer some type o tuition tax credit or deduction to assist amilies who want to send their children to independent

schools. Minnesota, Iowa and Illinois oer a direct tax credit or deduction to parents sending their children to private schools.Arizona, Florida, Pennsylvania, Iowa and Rhode Island oer credits to individuals or corporations that contribute to organizations

that provide private school scholarships. The experience o these states is directly relevant to the Georgia proposal.

By donating to scholarship organizations and receiving a tax credit in return, individuals and businesses contribute to

Georgias public good in an amount equal to what they would have paid had they not contributed to the scholarship organization.

Thus, total payment to the public good o Georgia by individuals and businesses is not lowered by the tax credit program; rather,

contributors to scholarship organizations ensure that their payments go directly to support the education o Georgia students. As

noted earlier in the study, unding contributed to scholarship organizations produces a larger overall increase in the educational

services purchased per dollar compared to money spent on state education aid.

In states such as Florida and Pennsylvania, the opportunity to direct tax payments to scholarship programs proved to bea powerul incentive or businesses to contribute, and in each state the initial caps placed on the total amount o business tax

credits were reached in the rst year o the program. Each state subsequently increased the total allowable tax credits.

Business Contributions to Scholarship Organizations

The experience o other states clearly indicates that we can reasonably expect businesses to contribute up to the maximum

amount allowed by the cap, $50 million, in the rst year. There are many reasons Georgia may want to provide a tax credit or

businesses that contribute directly to educating Georgias children. Doing so would:


22/36


2 February

Establish a convenient and consistent mechanism and incentive or businesses to contribute directly to educating

Georgias children.

Target educational expenditures directly to amilies and children rather than institutions that may reduce the amount

o resources that go directly to students.

Direct resources to students most in need o educational options and least likely to benet rom general increases in

school district budgets.

Give businesses a meaningul and convenient way to address their concerns about the quality o public education and

its impact on businesses and the Georgia economy.

Beyond the experience o other states, there are clear reasons why businesses can be expected to make contributions up to

the maximum tax credit allowance:

In nationwide surveys, nearly hal (48 percent) o all businesses indicated that education was their top priority or their

philanthropic and civic activities.15

In general, businesses are rewarded by owners, shareholders, equity analysts and nancial institutions or minimizing

the percentage o income going to taxes, because a lower tax burden is associated with sound nancial management

(even though contributions to Georgia tax-credit scholarships would be equivalent to the amount that would have been

paid in taxes).

Unlike tax payments, contributions to a program such as the Georgia tax-credit scholarship program are assumed to

generate concomitant civic and public relations opportunities and benets.

According to the Georgia Department o Revenue, 233,567 corporate income tax returns were led in Georgia in 2006,

with total taxable income o about $12 billion. Corporations paid about $812 million in corporate taxes to Georgia in 2006.16

Thesenumbers suggest that a tuition tax credit program that capped credits at $50 million would, at most, allow Georgia corporations

to oset about 6 percent o their corporate tax liability. However, because credits also are available to individuals as a credit

against personal income taxes, competition or available credits is likely.

Scholarship Contributions rom Individuals

According to data rom the Georgia Department o Revenue, i all $50 million in available credits were claimed by individuals

rather than corporations, the more than $8 billion in income tax liability or Georgia residents would be reduced by about 0.6

percent.17

To estimate the volume o contributions and tax credits that would be claimed by individuals we developed a model thatuses data on the charitable contributions o Georgia residents derived rom the Internal Revenue Services Statistics o Income;

historical survey research data on the percentage o charitable contributions that are directed to educational organizations; and

analyses o the experience o other states with tuition tax credits.

Among individuals, historically, about 20 percent o taxpayers have indicated that they contribute to educational organizations

and about 30 percent o their total contributions go to education organizations.18 Charitable contributions as a percentage o

Adjusted Gross Income (AGI) in Georgia have increased in recent years; in 2005, the most recent year or which contribution

data are available, Georgia residents claimed charitable tax deductions equal to just more than 3 percent o their total AGI, or

about $6.5 billion.19


23/36


February 2008

Total

Income

Projected Tax Credits Claimed by Individuals, by Taxpayer Income

Table 5

$1,326,806

$1,835,354

$5,523,450

$8,534,251

$6,753,570

$6,498,613

$2,749,932

$5,071,043

$2,299,564

$6,182,562

$46,775,147

2008

$1,366,611

$1,890,415

$5,689,154

$8,790,279

$6,956,177

$6,693,572

$2,832,430

$5,223,175

$2,368,551

$6,368,039

$48,178,401

2009

$1,407,609

$1,947,127

$5,859,828

$9,053,987

$7,164,862

$6,894,379

$2,917,403

$5,379,870

$2,439,607

$6,559,080

$49,623,753

2010

$1,449,837

$2,005,541

$6,035,623

$9,325,606

$7,379,808

$7,101,210

$3,004,925

$5,541,266

$2,512,796

$6,755,853

$51,112,466

2011

$1,493,332

$2,065,707

$6,216,692

$9,605,375

$7,601,202

$7,314,246

$3,095,073

$5,707,504

$2,588,179

$6,958,528

$52,645,840

2012

$1,538,132

$2,127,678

$6,403,193

$9,893,536

$7,829,239

$7,533,674

$3,187,925

$5,878,729

$2,665,825

$7,167,284

$54,225,215

2013

$1,584,276

$2,191,509

$6,595,288

$10,190,342

$8,064,116

$7,759,684

$3,283,563

$6,055,091

$2,745,800

$7,382,303

$55,851,971

2014

$1,631,804

$2,257,254

$6,793,147

$10,496,052

$8,306,039

$7,992,475

$3,382,070

$6,236,744

$2,828,174

$7,603,772

$57,527,530

2015

Arizona has more than eight years experience with an individual tax-credit scholarship program (more recently

Arizona has enacted a corporate tax-credit scholarship program). In Arizona, individual contributions to scholarship

organizations equal just over 1 percent o the total volume o charitable contributions in the state, and equal to about 0.03

percent o the total AGI o Arizona residents.20 We applied Georgias historical rate o charitable giving to projections o

the AGI o state residents to estimate the overall level o charitable contributions rom individuals in the state or each

year to 2015. We then applied ratios derived rom the Arizonas experience to produce two estimates (based on scholarship

contributions as a percentage o all charitable contributions in Georgia and scholarship contributions as a percentageo total AGI) o total contributions to student scholarship organizations. Our nal estimate was an average o the two

projections. Table 5 presents our estimate o contributions rom individual taxpayers by AGI in Georgia rom 2008 to

2015.21 The table indicates that individual taxpayers can be expected to claim tax credits or contributions in amounts

nearly equal to the capped amount, again suggesting that there would be signicant competition or credits between

corporations and individuals.

The gures in Table 5 represent the amount o contributions that can be expected in the absence o a cap on

available tax credits. Because contributions rom corporations can be expected to be much greater than those rom

individuals, the actual amount o contributions claimed by individuals would depend upon how credits are allocated

between corporations and individuals.

Tuition Prices Strongly Infuence Demand or Private Schools

The impact that a tax-credit scholarship program would have on public and private school demographics in

Georgia, as well as on state and local nances, depends on the dollar amount o contributions, the decisions o

scholarship organizations and the response o amilies o children in public and private schools to the availability o

scholarships. These are dicult to orecast. Program design elements and eligibility criteria would combine to infuence

the participation o Georgia amilies.

To estimate the response o Georgia amilies to the availability o tax-credit scholarships, we developed a model

o the demand or private schooling that allows the manipulation o key policy variables and program design elements.

Some o the variables included are:


24/36


4 February

The income-eligibility requirements or program participation;

The average dollar value o tuition scholarships; and

The expected price elasticity o demand or private schooling according to income level.

Tax-credit scholarships lower the price o private schools or students who receive them. A number o studies haveestimated the increase in demand or private schooling as a result o changes in the price o the schools. The most widely cited

studies o the impact o changes in the price o private schools on demand (the price elasticity o demand) indicate that the

demand or private schools increases as the price to amilies declines (and the demand decreases as the price rises), a so-called

negative price elasticity. The range o estimates between these studies is large, however. Chiswick and Koutroumanes (1996)

estimate a price elasticity o about -0.5, suggesting that a 10 percent decline in the price o private schools would lead to a 5 percent

increase in demand, while Gwarntey and Stroup (1997) estimate a price elasticity o -1.1, suggesting that a 10 percent decline in

the price o private schools would lead to an increase in demand o 11 percent.22 In Georgia, a 1994 study estimated the elasticity

o demand or private schooling in the state to be -1.07.23

Both the participation rate and scal impact o a scholarship program would be strongly infuenced by the dollar value

o the scholarships. To demonstrate the eect o changing the dollar value o scholarships, we consider a range o scholarship

values rom $1,500 to $4,500.

The number o scenarios and program design combinations is nearly innite. Our purpose is to create an understanding o

how design elements would aect program participation, and ultimately the scal impact. For Georgia amilies, a scholarship

with a value o $1,500 would represent a 25 percent reduction in the estimated average 2008 private school tuition o $5,940. 24 To

estimate program participation, we calculated the reduction in price that scholarships o various dollar values would have on our

orecast o the average price o tuition and applied dierent price elasticities and means tests to the distribution o school-age

children in public and private schools in Georgia.

Figure 5 presents our estimate o participation in a scholarship program in Georgia at dierent scholarship values i all

students were eligible regardless o income and i scholarships were o the same value regardless o amily income. The chart

shows that as many as 94,000 public school students, or about 6 percent, would seek to participate in a scholarship program with

these design eatures, while obviously all o the current private school students (not including those home schooled or in private

schools outside the state) would seek to take advantage o the program.

Figure 5 assumes that eligibility or scholarships is available to all children, regardless o amily income. However, i a

scholarship program were enacted in Georgia it might restrict eligibilityespecially or students currently enrolled in private

schoolsby means testing or in some other manner. For scal reasons that will be highlighted later in this study, it is benecial or

the state to make as many public school children eligible or scholarships as possible to encourage maximum migration rom the

public to private schools. Depending on the dollar value o scholarships, means testing or a reduction in the value o scholarships

as income rises can have a negative eect on the scal impact o a tax-credit scholarship program.

Figure 6 shows the impact on estimated demand or scholarships among public school students i eligibility is means

tested. The means tests are based on eligibility or the ederal ree and reduced-price lunch program as highlighted in Table 4.

Eligibility ranges rom the standard set or the ree and reduced-price lunch program (185 percent o ederal poverty guidelines)

to three times the amily income that would make a student eligible or ree and reduced-price lunches (about 555 percent o

ederal poverty guidelines).

The gure shows that means testing can dramatically reduce program participation because ewer public school amilies

are eligible. As important, the elasticity o demand or private schooling is much lower among lower-income amilies, meaning

they are less likely to participate in a scholarship program than higher-income amilies regardless o the value o the scholarship.


25/36


February 2008

140,000

120,000

100,000

80,000

60,000

40,000

20,000

0

Scholarship Demand Increases with Scholarship Value

Figure 5

$1,500 $2,00 $2,500 $3,000 $3,500 $4,000 $4,500

94,168

78,473

41,852

57,547

62,779

47,08452,315

73,24268,010

89,936

83,405

Demand by Current

Private School Families

Demand by Current

Public School Families

31,389

SCHOLARSHIP VALUE

STUDENTS

ESTIMATED DEMAND FOR TUITION SCHOLARSHIPS

100,000

80,000

60,000

40,000

20,000

10,000

0

Means Testing or Scholarship Eligibility Dramatically Impacts

Eligibility and Demand or Scholarships

Figure 6

$1,500 $2,000 $2,500 $3,000 $3,500 $4,000 $4,500

@ Free/Reduced Lunch Eligibility

2x Free/Reduced Lunch Eligibility

2.5x Free/Reduced Lunch Eligibility

3x Free/Reduced Lunch Eligibility

No Means Test

SCHOLARSHIP VALUE

DEMANDFO

RSCHOLARSHIPS

36,621


26/36


6 February

As we document in subsequent sections o this study, reducing eligibility among public school amilies actually results in lower

scal benets (or even scal losses) or the program. Thus, more restrictive means testing does not improve the scal impact

o a program.

On the other hand, restricting eligibility or participation among students currently attending private schools would yield

more scal benets to the state than i restrictions were applied to public school amilies. Because the decision to attend private

schools already has been made by those students, Georgia would receive no scal benet (in the orm o reduced state educationaid payments) rom increasing their eligibility. The primary eect o restricting the eligibility o current private school students

would be to reduce the competition or scholarships and increase the scal benets to the state.

That said, there is no justication or reducing or denying one group o citizens a benet that is available to others simply

because o where they chose to educate their children. This is especially true or lower-income amilies who may have made

tremendous sacrices by enrolling their children in private schools to obtain the educational services they believe are best or

their children. Nevertheless, restricting eligibility via means testing is an option in program design.

Figure 7 shows how eligibility or scholarships among private school students is aected by means testing based on multiples

o eligibility or the ederal ree and reduced-price lunch program. For example, i private school students participation is limited

to students rom amilies with incomes less than 2.5 times the level that would qualiy them or the ederal ree and reduced-price

lunch program, hal o private school students are eliminated rom eligibility, while lower-income amilies are not discriminated

against because they made sacrices to have their children educated in a school o their choosing prior to the enactment o a

scholarship program.

100%

67.0%

51.5%

40.1%

13.3%

0% 50,000 100,000 150,000 200,000

All

3x Free/Reduced Lunch

Eligible

Source: Author estimates based on U.S. Census Bureau American Community Survey 2006 data for Georgia

Only 13 Percent o Georgias Current Private SchoolStudents Qualiy or Free and Reduced Lunch

Figure 7

2.5x Free/Reduced LunchEligible

2x Free/Reduced Lunch

Eligible

Free/Reduced Lunch

Eligible21,934

66,126

84,901

110,432

164,911

CUMULATIVE # OF PRIVATE SCHOOL STUDENTS


27/36


February 2008

Combining Supply and Demand Models to Estimate the Number o Scholarships

The experience o states such as Arizona suggests that the number o scholarship applicants (i.e. demand) would be greater

than the available number o scholarships. In Georgia, the supply o scholarship money likely would be limited in each year, and

it likely would not be sucient to award scholarships to all applicants. As an example, we have estimated that the average private

school tuition in Georgia is $5,940 and a scholarship o $3,000 would reduce tuition by 51 percent on average. Using a reasonable

estimate o price elasticity o -0.75, a 51 percent decline in private school tuition should increase demand or private schools byabout 63,000 students currently enrolled in Georgia public schools, or about 32,000 i eligibility is limited to students rom amilies

with incomes below 2.5 times the income level or ree and reduced-price lunch eligibility. At the same time, scholarship unds

would be limited to $50 million, meaning that only 16,667 scholarships would be available. That is about 28 percent o the demand

or scholarships, or about 52 percent o demand i eligibility is restricted to children at 2.5 times ree and reduced-price lunch

eligibility. I scholarship unding were increased to $75 million, 25,000 scholarships would be available and 77 percent o the

demand or scholarships could be met.

Figure 8 highlights the relationship between the demand or scholarships and their supply at scholarship values ranging

between $1,500 and $4,500 and at several means-tested eligibility levels. These examples highlight the important relationships between

the total amount o scholarship money available; the average size o scholarship awards, program eligibility; and the migration

o students to private schools. These variables, along with the decisions o scholarship organizations, are dicult to model, butthey would determine the scal impact o the program. They are discussed more thoroughly in the ollowing sections.

Program Tradeos

100,000

80,000

60,000

40,000

20,000

10,000

0

Demand or Scholarships among Public School Families Increases with the Scholarship

Value, but the Supply o Scholarships Decreases

Figure 8

$1,500 $2,000 $2,500 $3,000 $3,500 $4,000 $4,500

Demand From Public School Students (Eligibility @ Free/Reduced Lunch Eligibility)

Demand From Public School Students (Eligibility @ 2x Free/Reduced Lunch)

Demand From Public School Students (Eligibility @ 3x Free/Reduced Lunch)

Supply o Scholarships @ $50 Million

Supply o Scholarships @ $75 Million

SCHOLARSHIP VALUE

NUMBEROFSCHOLARSHIPS


28/36


8 February

A tax-credit scholarship program could be constructed in various ways to yield important scal, educational, equity

and social objectives. A high-dollar-value scholarship does the most to attract low-income students to participate in a

choice program but would reduce the number o scholarships available. Conversely, relatively low scholarship values would

provide many more scholarships but would reduce overall program participation among the low-income amilies who need

educational options the most. Limiting participation to the lowest-income public school students would dramatically reduce

overall demand or scholarships and reduce scal benets while targeting the program to lower-income children.

These sorts o tradeos are inherent in all important public policies, and school choice programs are no exception.

The analyses and tools in this study are designed to make clear the impact o key program design variables and to highlight

the tradeos they imply.

Table 6 presents the cumulative distribution o projected scholarship demand by income or scholarship values ranging

rom $2,000 to $4,500. As has been noted, our analysis suggests that scholarships would induce a higher rate o public school

migration i the value o scholarships is increased and means testing or program eligibility is less restrictive.

Table 7 translates the estimates o scholarship demand in Table 6 into the context o the public school population and

shows that, at an average scholarship value o $4,500, about 6 percent o public school students would seek scholarships i

no income limits were established or scholarship eligibility. Lower scholarship values and means testing o eligibility would

have substantial impacts on scholarship demand.

185%

200%

250%

300%

Above 300%

Income Eligibility(% 0 Free/Reduced Lunch)

Table 6

12,525

37,760

48,480

63,059

94,168

$4,500

11,829

35,662

45,787

59,556

88,936

$4,250

11,133

33,564

43,094

56,053

83,705

$4,000

10,437

31,466

40,400

52,549

78,473

$3,750

9,741

29,369

37,707

49,046

73,242

$3,500

9,046

27,271

35,014

45,543

68,010

$3,250

8,350

25,173

32,320

42,040

62,779

$3,000

7,654

23,075

29,627

38,536

57,547

$2,750

6,958

20,978

26,933

35,033

52,315

$2,500

6,262

18,880

24,240

31,530

47,084

$2,250

5,56

16,78

21,54

28,02

41,85

$2,00

SCHOLARSHIP VALUE

Cumulative Scholarship Demand among Public School Students by

Scholarship Value and Income Eligibility

185%

200%

250%

300%

Above 300%


Cumulative Scholarship Demand among Public School Students

(as a Percentage o all Public School Students) by Income Eligibility

Table 7

0.8%

2.3%

3.0%

3.9%

5.9%

$4,500

0.7%

2.2%

2.8%

3.7%

5.5%

$4,250

0.7%

2.1%

2.7%

3.5%

5.2%

$4,000

0.6%

2.0%

2.5%

3.3%

4.9%

$3,750

0.6%

1.8%

2.3%

3.0%

4.6%

$3,500

0.6%

1.7%

2.2%

2.8%

4.2%

$3,250

0.5%

1.6%

2.0%

2.6%

3.9%

$3,000

0.5%

1.4%

1.8%

2.4%

3.6%

$2,750

0.4%

1.3%

1.7%

2.2%

3.3%

$2,500

0.4%

1.2%

1.5%

2.0%

2.9%

$2,250

0.3%

1.0%

1.3%

1.7%

2.6%

$2,00

SCHOLARSHIP VALUE


29/36


February 2008

Table 8 shows the cumulative percentage o scholarship demand that can be satised with tax-credit scholarships i $50

million o tax credits were allocated to scholarship organizations. The table shows that, i unds are limited to $50 million, a

ax-credit scholarship program could satisy demand or scholarships among public school students only i the average value

scholarships is relatively low (reducing demand and increasing the number o scholarships available) and more restrictive

ncome criteria are used. Unortunately, even in this situation the demand is met only because ewer public school amilies would

eek scholarships at such low values, and ewer would be eligible under the most restrictive income criteria. In addition, as wewill see later in this study, this combination o variables results in a scal loss or the state.

Table 9 shows the projected cumulative number o scholarships that would be awarded or each combination o eligibility

nd scholarship value criteria under a cap o $50 million. The table illustrates some subtle points about the design o a tax-credit

cholarship program. First, it suggests that, at average scholarship values o $4,000 or less, it would be important to increase the

ncome-eligibility criterion above the level that qualies a student to receive ree and reduced-price school lunches. I the income

riterion is not above the ree and reduced-price lunch eligibility threshold, there likely would be insucient demand amongublic school students to use all the $50 million in scholarships. One implication o that is that Georgia would not maximize its

scal benets, a point that is demonstrated in the next section.

The table also illustrates the key points that amilies with higher incomes would participate at rates higher than those o

ower income even with lower scholarship values, and the demographic mix o participants would shit more to higher-income

185%

200%

250%

300%

Above 300%


Percentage o Scholarship Demand among Public School

Students that Can Be Satisfed by $50 Million in Scholarships

able 8

88.7%

29.4%

22.9%

17.6%

11.8%

99.5%

33.0%

25.7%

19.8%

13.2%

112.3%

37.2%

29.0%

22.3%

14.9%

127.7%

42.4%

33.0%

25.4%

17.0%

146.6%

48.6%

37.9%

29.1%

19.5%

170.1%

56.4%

43.9%

33.8%

22.6%

199.6%

66.2%

51.6%

39.6%

26.5%

237.5%

78.8%

61.4%

47.2%

31.6%

287.4%

95.3%

74.3%

57.1%

38.2%

354.9%

117.7%

91.7%

70.5%

47.2%

449.1%

149.0%

116.0%

89.2%

59.7%

$2,000$2,250$2,500$2,750$3,000$3,250$3,500$3,750$4,000$4,250$4,500

the fiscal impact of tax-credit scholarships in georgia

Documents